7

I am in desperate need of help to improve the abysmal performance of my 802.11n wireless network. At best I get 30Mbs (this is an internet download) from a technology that boasts 300Mbs, even worse is the LAN where to date the best I have ever gotten is 1Mbs. It is literally quicker to copy the file to a USB drive and walk it to the other computer.

Infrastructure is this

  • AP 802.11n only broadcasting at both 2.4GHz and 5GHz
  • Mac with 802.11a/b/g/n card is connected to the AP via 5GHz
  • Linux with 802.11a/b/g/n card is connected to AP via 2.4GHz

I have conducted the following tests (results at end of post)

  • Internet based speed test wired and wireless
  • LAN file copy wired and wireless

I have read:

I have made the following optimizations

  • AP broadcasts only 802.11n on both 2.4GHz and 5GHz frequencies
  • 2.4GHz is on a channel with least interference (live in an apartment with lots of APs), this did make a 10Mb/sec improvement
  • Our AP is the only one transmitting on the 5GHz freq.
  • Security: WPA Personal WPA2 AES encryption
  • Bandwidth: 20MHz / 40MHz (I assume this to be channel bonding)

I have tried the following with 0 improvement

  • Dropped the Fragment Threshold to 512
  • Dropped the Request To Send (RTS) Threshold to 512 and 1
  • Even thought of buying a frequency spectrum analyzer, until I saw the cost of them!!!

Speed test results

Linux Wired:

Mac Wired:

Linux Wireless:

Mac Wireless:

LAN NFS 53,345,087 bytes (51MB) file

  • Linux Mac NFS Wired: 65.6959 Mb/sec
  • Linux Mac NFS Wireless: .9443 Mb/sec

All help is appreciated, even testing methods will be accepted.

2
  • 1
    None of those speeds seem to be concering. What makes you think your access point has the ability to provide faster speeds? Your wire and wireless upload speeds match, this means, your using 100% of the bandwidth and have very little packet loss. One has to remember that the 54 Mbit/s to 600 Mbit/s are the speeds you would see in a clean room. They are also an entirely different unit of measurement.
    – Ramhound
    Sep 6, 2013 at 11:30
  • How many Access Points can your computers see ? You will probably find that the Wifi provider is lying to you about speed (they ALL do it, its pretty much written in to the spec as you will never get anything close to the advertised speed in real life), and the amount of other devices "eating" into your available spectrum. You might be able to speed things up by changing WIFI channels, but if you can't you can always kill the power to your neighbours so their WIFI's stop interfering with yours :>
    – davidgo
    Sep 7, 2013 at 4:45

8 Answers 8

1

Sometimes you may have other devices using 2.4Ghz causing interference. Not sure if you have tried using a scanner to see if some other device is interfering. I've found this to be the case at my job where I had a similar situation and there was a telephone using 2.4Ghz that caused slowdowns.

1
  • Normally N wireless runs on 5Ghz, but in this case, it's running mixed mode with a/b/g, so yes, this is a good idea. Lots of random devices crap all over the 2.4Ghz including phones, baby monitors, and microwaves.
    – Karu
    Mar 18, 2016 at 19:54
0

What is serving the NFS data? Is this on the wired network?

See http://www.tp-link.us/article/?faqid=190 for finding the MTU fragmentation limits. If you can set the MTU on whatever is serving your NFS then you might be able to improve the wi-fi performance.

0

If your wireless router is set to both 20/40MHz auto selection for Channel Bandwidth and there are a number of other Access Points in the area it may be having trouble using enough channels to actually support the 40MHz option.

When using the 20MHz option only one channel is used in the transmission of data, and makes it very obvious if you use any channel scanning utility on a computer or phone that it is overlapping with other Access Points in close proximity.

When attempting to work at 40MHz the router is actually extending from the base channel either up or down (normally this can be configured).

I followed advice for some time to simply using 20MHz channel bandwidth exclusively, but later was able to once again use the 20/40Mhz auto setting by being extremely particular about my control channel choice and the direction of channel extension.

I used the following application on my Android phone to check the impact of other Access Points on the channels I attempted to use:

Wifi Analyzer (Google Play)

0

I recently resolved issues (dropouts, slow speed, etc) with my 802.11n wireless network by switching to one of the newer 'beamforming' routers. Since switching, everything is stable and fast now. :) I was previously using a good quality Linksys router, but it couldn't handle the interference from multiple nearby devices and walls, etc.

Switching to a beamforming router might be worth a try if nothing else works.

Cheers

0

For the Internet, I have to agree with RAMHound that those speeds are about normal. Wireless has a lot more overhead so actual transfer rates are much lower than the spec. One thought though is the wireless speeds look about right for the original (draft) 150 Mbit 802.11 N. If you only upgraded the router/AP to 300N then you probably have the 150N cards in the computers or they are set to single channel mode (see http://compnetworking.about.com/od/wireless/f/80211n-300-mbps.htm).

For the LAN NFS, the difference here is much greater. Based on the Internet tests, on WiFi you should have about 12 Mbits. If you only tested that once or twice, it could have been a temporary issue otherwise the NFS may need optimizing for WiFi. This might help: http://ubuntuforums.org/archive/index.php/t-1047907.html

For testing the LAN easily, I use LAN Speed Test which has a Mac version available:

http://www.totusoft.com/downloads.html

0

This sounds like an RF issue, whose only solutions are (choose 1):

  1. Move
  2. Run ethernet cabling to all your devices

We can't solve RF issues with software. Its a hardware issue.

0
0

From your description, your problem is not software related, its hardware related. More properly, its a radio propagation problem.

I am not sure if it is a problem at all. Those speeds are about right for the setup.

First: WiFi networks are half-duplex. This means that the neat numbers you see on the boxes of the Access Points at the market are not exactly the running speeds of those devices. By Half-duplex this means that the quoted speeds are one way only. Its the signalling speeds of the channel. To test this you might put two access points (one as AP the other as client) into a anechoical chamber and start transfering data both ways. What happens ? You wont see anywhere near the 150mbit/sec hated, you will see something close to HALF the speed, because the channel is shared between transmiting and receiving. So, it would be more like 75mbit/sec.

Second: The values advertised are - conveniently - measured in bits per second. But what you transfer between computers is usually bytes. This means that to have a figure in bytes per second you have to divide the values by eight. And this is not enough. Every time your computer wants to communicate with the access point and vice versa, this is done via PACKETS, packets are limited ammounts of data that are marked with addresses and flags. This means that for each certain ammount of data you transfer between your computer and your access point, a certain ammount of boiler plate information will have to be transfered between them. So make those 75mbit/sec up/down at the same time be more like 40mbit/sec up/down.

Third: You are not alone in your channel. Even if you have no known nearby source of emission of radiation on the same channel, eletromagnetic radiation is generated by a lot of physical processes in our own world. When a radio wave is not transmitting information, its transmitting noise. The most common source of noise is heat. No, your access point is not getting hot and generating noise. Everything around us has a temperature. Be it 100F or 20C, any temperature above the absolute zero will produce radio waves. We call that radio waves "thermal noise". On any channel theres something called noise floor, thats the lowest ammount of noise present on that channel. On top of the natural noise, nearby access points - even if present in other channels - raise the noise floor, because theres no perfect way to "insulate" one channel from the other. And the more close they are to each other, the more powerfull this radiation becomes. The capacity of an access point to transmit information using radio waves is limited by what we call "signal to noise ratio". If there is more power in the signal, this value grows, if there is more power in the noise, this value falls. If you increase your signal too much, instead of generating a signal with a better signal to noise ratio, you might very well increase your noise floor due to the components in the signal path of the radio operating at a higher temperature where they generate more thermal noise. So, you must find the correct value for the power setting in your device (and your neighboor incorrectly configured AP might interfere here too). That being said, the ultimate arbitrer of the speeds that you might get out of your access point is the noise floor that all the devices see in the selected channel. Walls in the path of the signal might attenuate it too much. Bad placement of antennas on the devices themselves etc. Everything can be a reason for a lower performance. So the previous 40mbit/sec might be 20mbit/sec, for example.

Forth I advise you not to change RTS/Fragment threshold values. RTS is not very usefull in a "soho" environment. This value is used to decide if its worth asking for a CTS (clear to send) before sending (a process that takes some time but might prevent collisions) or sending the packet straight on. If the packet is smaller than the value of RTS threshold it will be sent straight away, if its equal or bigger it will grant a RTS/CTS handshake. This is done to avoid the hidden node problem. But for shorter ranges, all devices should be, usually, within sight of each other. The fragment threshold is used to decide when to fragment TCP/IP packets into smaller frames. This might be usefull under heavy interference ambients, when all channels are occupied and losing a big packet to collision is worse than losing a series of small packets. It might be usefull when you have a non-wifi source of interference too (i.e.: repetitive interference that might consistently drop big packets). I dont believe this is your case.

What can be done ? Take a screenshot of your access point web management screen, on the tab that shows all the connected clients so i can check out whats happening more profoundly. The 802.11n protocol specifies various modes of connection (MCS) to take most out of a channel, and each of those modes has its own characteristics speeds, lags etc. So the first thing to consider is the MCS mode that is being used by the devices.

2
  • Pls, who gave the downvote, specify why. It would be very helpful, i would like to improve my answer.
    – Jorge Aldo
    Mar 27, 2015 at 4:27
  • Upvoting to balance to 0. Lots of good info in the post, no obvious reason why it should be downvoted. Oct 12, 2015 at 22:55
0

You say there are many other APs in your vicinity.

It is to be noted that if there are other wireless networks and devices on the same channels as your's Specifically 802.11b. When repeated unsuccessful frame transmissions are detected. Hosts Will lower their Bitrates** at A, B or G speeds Or even lower, depending on the speed of the Slowest Hosts of that wireless Channel.

Reason: When mobile hosts move, they may encounter bad transmission conditions and degrade the bit rate from 11 Mb/s to 5.5, 2, or 1 Mb/s. A host with a lower bit rate will influences the throughput of other hosts that share the same radio channel.

When repeated unsuccessful frame transmissions are detected. In such a case, a host transmitting for example at 1Mb/s reduces the throughput of all other hosts transmitting at 11Mb/s to a low value below 1 Mb/s. The basic CSMA/CA channel access method is at the root of this anomaly: it guarantees an equal long term channel access probability to all hosts.

Our analysis and performance measurements show that the slow host may considerably limit the throughput of other hosts roughly to the level of the lower rate.

Solution: Try selecting a channels that are is only N or make the appropriate changes on those other (Older) routers/Networks to free channels for the N standard.

source:Here is the paper explaning exactly how and why it happens

2
  • This is not true. The presence of other APs using B/G or A will not interfere on this AP hability to use mode N
    – Jorge Aldo
    Mar 27, 2015 at 3:51
  • I changed my answer To explain why, Other Devices and/or network Using the same Channel Can degrade The speed of all Devices using That channel, I Added The source of a Published papers explaining why.
    – GuruJR
    May 24, 2015 at 1:58

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .